The present invention relates to computing environments including shared resources, and more specifically, this invention relates to maintaining the integrity of a shared volume.
In environments where a shared resource, such as a storage volume, is accessible to multiple host systems, some host-initiated operations may transform the resource in a fundamental way. When one of these operations is performed by one of the connected host systems, it may be necessary to render the shared resource unavailable to the other connected host systems. In other words, some action may be performed to make the shared resource unavailable for use to applications on the other connected host systems. In one particular example, on a host system running z/OS, a VARY command may be issued to vary the shared resource offline to z/OS.
If the shared resource remains available to one or more host systems during the operation, then data corruption or undetected data loss can occur.
In some use cases, such as Hierarchical Storage Management (HSM) Fast Replication Backup, a shared volume may remain online because it is expensive to vary numerous devices offline/online to a host system that will then back up the shared volume to tape.
Additionally, after performance of the operation completes, the shared resource may then again be made available to the host systems. If a host system that is unaware of the operation attempts to write to data sets that have been moved as a result of the operation, then the effects of the operation may be lost, and/or a file directory may become corrupted.